Inter-layer Video Disparity Vector Prediction via Spatial Proximity

The patent applies local quality by prioritizing neighboring blocks based on their spatial proximity to the current block. Specifically, it defines a search order where blocks closer to the current block (e.g., left, top, bottom neighbors) are examined first, while farther blocks are examined later or skipped. This localized approach ensures that the most relevant neighboring blocks are selected for disparity vector prediction, improving accuracy without requiring a complete search of all possible blocks.

The patent segments the neighboring block search space into priority levels based on distance from the current block. Instead of treating all neighboring blocks equally, it divides them into groups (e.g., immediate neighbors vs. distant neighbors) and processes them in a specific order. This segmentation allows the system to focus computational resources on the most promising candidates first, achieving better prediction accuracy with reduced complexity.

The patent emphasizes local quality by giving higher priority to neighboring blocks that are spatially close to the current block. The search algorithm examines blocks in order of their proximity, ensuring that locally relevant blocks are selected as candidates before considering distant blocks. This approach maintains reliability by focusing on the most pertinent local information while still allowing broader search coverage if needed.

The patent applies preliminary action by performing a prioritized search where close neighboring blocks are examined first before distant ones. The algorithm establishes a predetermined search order based on spatial distance, so that the most likely candidates for accurate disparity vector prediction are evaluated early in the process. This preliminary focus on high-probability candidates improves selection accuracy while maintaining efficient search coverage.